Structural equation modeling of the Restructured Clinical scales of the MMPI-2

The Restructured Clinical scales of the MMPI-2 represent a major revision of the Clinical Scales. In deriving the new scales, Tellegen et al. (2003) adopted an exploratory approach to test construction utilizing flexible criteria and clinical judgment to augment empirical data.;The aim of this investigation is to further validate the Restructured Clinical scales by assessing the fit of items to their respective scales, identifying items that do not fit well, and illuminating the nature of the scale correlations by identifying latent factors in the scales. This information will assist users in interpreting the new scales and provide information that will potentially aid others in the refinement of the scales.;The item-scale analyses were performed using confirmatory factor analysis. RCd Demoralization, RC1 Somatic Complaints, RC3 Cynicism, RC6 Ideas of Persecution, RC7 Dysfunctional Negative Emotions, and RC8 Aberrant Experiences show good fit with the sample data while RC2 Low Positive Emotions, RC4 Antisocial Behavior, and RC9 Hypomanic Activation show poor fit with the sample data. Specific items that do not fit well with their respective scales are identified.;Inter-scale analyses were performed using confirmatory factor analysis and structural regression modeling. A hierarchical model in which shared variance from Demoralization accounts for correlations between Somatic Complaints, Low Positive Emotions, and Dysfunctional Negative Emotions and shared variance from Dysfunctional Negative Emotions account for many of the correlations between the other scales produces the best overall fit with the sample data. Evidence for hostility-dyscontrol and psychotic factors in some of the scales are also presented. Implications and limitations are also presented.;Although such an approach is wholly appropriate for test construction, it raises questions that should be resolved in order to help clinicians and researchers in interpreting the new scales. For example, in assigning items to the scales, Tellegen et al. (2003) adopted low minimum item-scale correlation criteria that represent threats to the internal consistency of the scales. Additionally, although they sought to remove a general factor of psychopathology from the Clinical Scales that they believe is responsible for scale correlations, their validation research shows that meaningful correlations persist in the Restructured Clinical scales.